I'm currently doing shadowing in order to do measurements for a couple of activities/processes that I've mapped before this. I would like to know what's the best method to calculate the number of instances that I need to measure for every activity/process in order to be able to calculate an accurate average processing time per activity/process?
Thanks in advance for your answers!
the typicall number of measurements I do is 30 but this can vary depending on the consistency of the results and the deviations I get.
If 15-20 measurements are grouped close to each other and I do not see any signifficiant deviations from one measurement to another I am quite happy with this sample and I do not go further. If this is not the case I go for 30 and sometimes even more and I eliminate the 5-10% outliers that might appear but very important I would also try to clasify this cases and understand the route causes of their manifestation.
Also it dependends on your stakeholders. If you feel that there will be challanging discussions about the times and your worked could be questioned because of this maybe it's healthier to agree with your internal customer what is the sample that will be considered relevant by both parties envolved.
All the best,